|Publication number||US4700229 A|
|Application number||US 06/879,697|
|Publication date||Oct 13, 1987|
|Filing date||Jun 27, 1986|
|Priority date||Jul 9, 1985|
|Also published as||DE3686848D1, DE3686848T2, EP0209180A1, EP0209180B1|
|Publication number||06879697, 879697, US 4700229 A, US 4700229A, US-A-4700229, US4700229 A, US4700229A|
|Inventors||Jochem Herrmann, Marcellinus J. M. Pelgrom|
|Original Assignee||U.S. Philips Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (2), Non-Patent Citations (4), Referenced by (44), Classifications (13), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
A(1). Field of the Invention
The invention relates to an image enhancement circuit. Such circuits are used to produce a high quality, image from a plurality of quasi-stationary low quality images. For example, let it be assumed that a high-quality TV image is desired of an object which is at a long distance from a TV camera. When the atmosphere is quiet (as with light frost), a high quality image can generally be obtained without image enhancement. However, if the atmosphere is very turbulent (as on very hot days) a low quality is generally produced, with the image being highly degraded by noise. To obtain an image of high quality under such circumstances, a plurality of low quality images of this object is processed in the image enhancement circuit.
Similar situations occur in scanning electron microscopy, X-ray fluoroscopy and in astronomy.
A(2). Description of the Prior Art
A known method of obtaining an image of high quality from a series of low-quality images is described on page 434 of a book by W. K. Pratt entitled Digital Image Processing (John Wiley and Sons, 1978). In this method, the average value is of the series of low-quality images is calculated. This may lead to the desired result, which is evident as follows. For the i-th low-quality image Gi (x,y),
Gi (x,y)=H(x,y)+Ni (x,y) (1)
In this equation H(x,y) represents the desired high-quality image and Ni (x,y) represents an additive noise image which is independent of the high-quality image. When the desired high-quality image is constant for a series of M low-quality images, ##EQU1## For a high value of M, the noise term on the right-hand side approximates zero under normal circumstances, so that: ##EQU2##
A device for performing the above-mentioned method will generally comprise a circuit for scanning the low-quality images Gi (x,y) in an image scan period Tb, for generating low-quality image signals gi (t). These signals are weighted in a weighting circuit with a weighting factor 1/M. The low-quality image signals thus weighted are accumulated in an accumulator. After all M low-quality image signals have been applied to this accumulator, the supply of further low-quality image signals is impeded. At that moment, the output of the accumulator supplies a high-quality image signal f(t) which represents the desired high-quality image H(x,y). The high-quality image can be displayed by applying f(t) to a monitor.
As has been described on page 686 of an article by H. Bruggemann entitled "Temporal Filtering Using Pixel Incrementing" (Journal of the Society of Motion Picture and Television Engineers, pages 686-694 (August, 1981), a drawback of this known image enhancement circuit is that the desired brightness of the image cannot be obtained until after all M weighted low-quality image signals are accumulated. A further drawback of this known circuit is that the optimum value of M is not always known in advance. In fact, this will be greatly dependent on the circumstances.
It is an object of the invention to provide an image enhancement circuit in which a high-quality image signal is continuously obtained.
To this end, the image enhancement circuit according to the invention further comprises a difference-producing circuit to which the low quality signals as well as the output image signals occurring at the output of the accumulator are applied. An output image signal is subtracted from each low-quality image signal to generate difference image signals which are applied to the weighting circuit in which these difference image signals are weighted by a weighting factor of k(i). The magnitude of K(i) is dependent on the ordinal number of the relevant difference image signal.
By assuming the weighting factor for the i-th difference image signal to be equal to 1/i, the output signal of the accumulator continuously represents the average of all low-quality image signals applied up to that moment to this image enhancement circuit so that the image on the monitor continuously has an optimum brightness. In addition the accumulation process may now be interrupted at any desired moment.
In a preferred embodiment of the image enhancement circuit according to the invention, the weighting factors are chosen to be equal to integral negative powers of two, particularly as follows:
k(i)=2-RND[.spsp.2log i] ( 4)
In this equation RND represents a rounding-off operation; in other words RND[2 log i] is the rounded-off value of 2 log i. The weighting factors determined in this manner successively have the value
1, 1/2, 1/4, 1/4, 1/4, 1/8, 1/8, 1/8, 1/8, . . . (5)
It is true that the image enhancement circuit throretically does not yield the same image quality when k(i)=1/i when the latter weighting factors are used, but in practice the difference in quality can hardly be ascertained, also because the accumulation process is not to be stopped after a given period, the weighting circuit can be cnsiderably simplified by this choice of the weighting factors.
FIG. 1 schematically shows a television recording and display system provided with an image enhancement circuit according to the invention.
FIGS. 2 and 3 schematically show embodiments of portions of an image enhancement circuit according to the invention.
FIG. 1 shows a TV system. The system comprises a TV camera 1 which converts a series of images Gi (x,y) of one and the same objects at a rate of, for example, 25 Hz into a series of analog image signals gi (t). Reference i represents the ordinal number of the image signal. For this conversion, the image Gi (x,y) is scanned in a line pattern.
After the image Gi (x,y) is scanned and the image signal gi (t) is obtained, the image will have changed slightly. This new image will be indicated by Gi+1 (x,y). After scanning by the TV camera, the camera yields the image signal gi+1 (t). In addition to these image signals, the TV camera supplies a frame-synchronizing pulse fp whenever scanning of a new image is started.
The analog image signals gi (t) thus obtained are converted in a circuit 2 into time-discrete image signals gi (n). To this end, this circuit 2 includes a sampler 21 which is controlled by sampling pulses at a rate fs of, for example, 13.5 MHz. In the embodiment shown, this circuit 2 further includes an analog-to-digital converter 22 in which each image signal sample thus obtained is converted into a code word of, for example, 7 bits.
The time-discrete image signals which are supplied by the circuit 2 are applied to the input 31 of the image enhancement circuit 3. The output 32 thereof is connected to the monitor 5 via a digital-to-analog converter 4.
The image enhancement circuit 3 has an accumulator 33. In response to the i-th image signal gi (n), the accumulator 33 yields the i-th approximation hi (n) of the desired image signal h(n) which represents the image H(x,y).
The accumulator 33 has an input 331, an output 332, and an adder circuit 333. Adder 333 has a first input connected to the input 331 of the accumulator and a second input connected to the output 332 of the accumulator. Accumulator 33 also has a delay circuit 334 having a delay time Tb which is equal to the duration of an image signal. The input of delay circuit 334 is connected to the output of the adder circuit 333. The output of delay 334 is connected to the output 332 of the accumulator.
The image enhancement circuit further includes a difference producing circuit 34. A first input of difference circuit 34 is connected to the input 31 of the image enhancement circuit. A second input of difference circuit 34 is connected to the output 32 of the accumulator 33. The output of this difference producing circuit 34 is connected via the weighting network 35 to the input 331 of the accumulator.
The weighting network 35 provides a weighting factor k(i) dependent on the ordinal number i of the low-quality image signal. With the aid of a switch 36, which can be operated by the user, the weighting factor can be set to a given starting position and can be maintained in this position, if desired, by leaving this switch permanently operated.
It will be assumed that the weighting factor at the start is equal to k(0)=0. Since the first synchronizing pulse fp occurs after switch 36 has opened, i becomes 1 and, starting from the assumption that k(i)=1/i, the first weighting factor becomes equal to k(1)=1. The image signal gi (n) is then stored unchanged in the delay circuit 334.
After the next frame synchronizing pulse fp, the stored image signal is input together with g2 (n) to the difference producing circuit 34. Circuit 34 then supplies the difference signal g2 (n)-g1 (n). This difference signal is multiplied by the new weighting factor 1/2, and is applied to the adder circuit 333 where the image signal g1 (n) is added to this difference signal. The sum signal
s2 (n)=1/2[g1 (n)+g2 (n)]
thus obtained is applied to the delay circuit 334. After the next frame synchronizing pulse, s2 (n) is applied as an image signal h2 (n) to the monitor 5 via the digital-to-analog converter 4.
As can be easily derived from the above, it generally holds that ##EQU3## When the quality of the image Hi (x,y) corresponding to hi (n) and visible on the monitor 5 is considered to be sufficient by the user, switch 6 can be closed again so that the image Hi (x,y) does not change anymore. As is evident from equation (6), hi (n) is the average value at any moment of all image signals g1 (n) received until that moment, so that the image Hi (x,y) has an optimum brightness.
An embodiment of the weighting circuit 35 in the case where k(i)=1/i is shown in FIG. 2. It comprises a counter 3501 having, for example, 16 bits. The frame synchronizing pulses fp are applied to the counting input C. This counter can be reset, for example, by applying a logical "1" to the reset input R. For this purpose switch 36 is provided which is operated by the user.
The counter thus supplies the ordinal number i of image signal gi (t) which is applied to the image enhancement circuit since switch 6 was last closed. This number i, together with the difference signal provided by the difference producing circuit 34, is applied to a divider circuit 3503. Divider 3503 thus supplies the difference signal weighted by the weighting factor 1/i.
It is to be noted that the 216 frame synchronizing pulses that can be counted by this 16-bit counter correspond to 65536 images. Transmission thereof takes approximately 43 minutes. Therefore, in this case the user must determine within approximately 43 minutes whether there is an image satisfying his wishes.
In the weighting circuit shown in FIG. 2, k(i)=1/i and therefore the circuit provided with a divider circuit 3503. Although divider circuits are generally known, they remain relatively complicated circuits. In addition it takes a relatively long time to carry out a division.
A considerably simpler embodiment of the weighting circuit is obtained by choosing the weighting factors to correspond to equation (4) which leads to the series of weighting factors indicated in equation (5). This choice of the weighting factors not only considerably simplifies the weighting circuit design, but in addition considerably increases its speed.
An embodiment of the weighting circuit in which the weighting factors have the values of equations (4) and (5) is shown in FIG. 3. In this embodiment it has been assumed for the sake of simplicity that both the difference signal samples, which are supplied by the difference producing circuit 34, and the weighted difference signal samples, such as are applied by the weighting circuit, are all represented in sign and magnitude and are located between +1 and -1.
The weighting circuit shown in FIG. 3 again comprises the 16-bit counter 3501 which receives the frame synchronizing pulses fp at its counting input C and which can be reset by applying a logical "1" to the reset input R for a short period. Here again switch 36 is provided which is operated by the user, but it only conducts for a short period.
A decoding network 3504, having sixteen outputs indicated by S1 to S16, respectively, is connected to the counter 3501. A logical "1" occurs at output S1 when the count is one. If the count is two, a logical "1" occurs at the output S2. A logical "1" occurs at the output S3 when the count is 3, 4, and 5. Thus, in general a logical "1" appears at output Sn, where n=RND(2 log i)+1, when the count is i. (See, equations (4) and (5).)
The weighting circuit further includes a 23-bit output register 3505 whose contents (being the weighted difference signal samples) are transmitted in parallel to the adder 333 (see FIG. 1). This output register has a bit cell for each of the 23 bits. These cells are indicated by BS, B1, B2, . . . B22, respectively. Bit cell BS comprises the sign bit, bit cell B1 comprises the most signigicant bit MSB and bit cell B22 comprises the least significant bit LSB.
Of each 7-bit difference signal sample which is supplied by the difference producing circuit 34, the sign bit is directly applied to bit cell BS. The other six magnitude bits are applied via a routing circuit 3506 to certain other bit cells of the output register 3505. To this end, this routing circuit includes sixteen gating circuits 3507-3522. Each gating circuit has six inputs for receiving the six magnitude bits of the difference signal in parallel. For the sake of completeness FIG. 3 shows at which of the gating currents' outputs the MSB is present and at which the LSB is present.
The gating circuits are controlled by the signals at the outputs of the decoding network 3504. More particularly the six magnitude bits are applied to the bit cells B1-B6 of output register 3505 when output S1 has a logical "1". If on the other hand output S2 has a logical "1", these six magnitude bits are applied to the bit cells B2-B7 (thus dividing by 2). When output S3 has a logical "1", they are applied to the bit cells B3-B8 (thus dividing by 4). Thus, in general, the six magnitude bits of the difference signal are applied to bit cells Bn-B(n+5) of output register 3505 when output Sn is a logical "1". The gating circuits are designed in such a manner that gating circuits receiving a logical "0" do not receive any of the six magnitude bits.
It is to be noted that it has been assumed for the sake of simplicity in this embodiment that the difference signal samples are represented with sign and magnitude. The twos-complement representation, however, is generally used in practice.
It is also to be noted that the time delay element 334 (see FIG. 1) may alternatively be a digital image memory.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4058836 *||Apr 26, 1976||Nov 15, 1977||The Marconi Company Limited||Noise reduction in television signals|
|US4607284 *||May 9, 1984||Aug 19, 1986||U.S. Philips Corporation||Movement-adaptive transversal-recursive noise suppression circuit for a television signal|
|1||*||Bruggemann, Harro., Temporal Filtering Using Pixel Incrementing. J. Soc. Motion Picture and Television Engineers, pp. 686 694 (Aug., 1981).|
|2||Bruggemann, Harro.,"Temporal Filtering Using Pixel Incrementing." J. Soc. Motion Picture and Television Engineers, pp. 686-694 (Aug., 1981).|
|3||*||Pratt, W. K., Digital Image Processing. John Wiley and Sons, pp. 434 437 (1978).|
|4||Pratt, W. K., Digital Image Processing. John Wiley and Sons, pp. 434-437 (1978).|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US4829380 *||Dec 9, 1987||May 9, 1989||General Motors Corporation||Video processor|
|US4926361 *||Sep 16, 1988||May 15, 1990||Hitachi, Ltd.||Digital noise reducer|
|US4985618 *||Jun 13, 1989||Jan 15, 1991||Nicoh Company, Ltd.||Parallel image processing system|
|US5001663 *||May 3, 1989||Mar 19, 1991||Eastman Kodak Company||Programmable digital circuit for performing a matrix multiplication|
|US5045952 *||Aug 21, 1989||Sep 3, 1991||Xerox Corporation||Method for edge enhanced error diffusion|
|US5134503 *||Jul 19, 1990||Jul 28, 1992||Fuji Photo Film Co., Ltd.||Image processing apparatus|
|US5201013 *||Apr 20, 1990||Apr 6, 1993||Ezel, Inc.||Dither processing method|
|US5289283 *||Aug 9, 1990||Feb 22, 1994||British Telecommunications||Inter-frame video filter having selective linear and non-linear transfer characteristics as function of inter-frame differences|
|US5315669 *||May 8, 1992||May 24, 1994||Ezel Inc.||Dither processing method|
|US5325211 *||Jan 4, 1993||Jun 28, 1994||Xerox Corporation||Error diffusion with output and input based feedback|
|US5418563 *||Dec 6, 1993||May 23, 1995||Panasonic Technologies, Inc.||Dynamic peaking aperture correction for use with a CCD camera|
|US5424783 *||Feb 10, 1993||Jun 13, 1995||Wong; Yiu-Fai||Clustering filter method for noise filtering, scale-space filtering and image processing|
|US5438634 *||May 23, 1994||Aug 1, 1995||Ezel Inc.||Dither processing method|
|US5493416 *||Oct 31, 1994||Feb 20, 1996||Xerox Corporation||Method combining error diffusion and traditional halftoning with arbitrary screen orientation|
|US5621546 *||Nov 2, 1993||Apr 15, 1997||Xerox Corporation||Method and apparatus for vector error diffusion with output color control|
|US5668638 *||Jun 27, 1996||Sep 16, 1997||Xerox Corporation||Error diffusion method with symmetric enhancement|
|US5859931 *||May 28, 1996||Jan 12, 1999||Xerox Corporation||Image compression and decompression using predictive coding and error diffusion|
|US6380539||Jan 23, 1998||Apr 30, 2002||Applied Science Fiction, Inc.||Four color trilinear CCD scanning|
|US6393160||Mar 9, 1999||May 21, 2002||Applied Science Fiction||Image defect correction in transform space|
|US6437358||Feb 4, 1999||Aug 20, 2002||Applied Science Fiction, Inc.||Apparatus and methods for capturing defect data|
|US6442301 *||Dec 29, 1997||Aug 27, 2002||Applied Science Fiction, Inc.||Apparatus and method for defect channel nulling|
|US6487321||Sep 15, 2000||Nov 26, 2002||Applied Science Fiction||Method and system for altering defects in a digital image|
|US6498867||Oct 10, 2000||Dec 24, 2002||Applied Science Fiction Inc.||Method and apparatus for differential illumination image-capturing and defect handling|
|US6590679||Jan 26, 1999||Jul 8, 2003||Applied Science Fiction, Inc.||Multilinear array sensor with an infrared line|
|US6593558||Dec 30, 1999||Jul 15, 2003||Applied Science Fiction, Inc.||Luminance-priority electronic color image sensor|
|US6614946||Oct 5, 2000||Sep 2, 2003||Eastman Kodak Company||System and method for correcting defects in digital images through selective fill-in from surrounding areas|
|US6650789||Jan 18, 2002||Nov 18, 2003||Eastman Kodak Company||Method and system for altering defects in a digital image|
|US6683995||Dec 12, 2000||Jan 27, 2004||Eastman Kodak Company||Method and apparatus for correcting large defects in digital images|
|US6704458||Dec 12, 2000||Mar 9, 2004||Eastman Kodak Company||Method and apparatus for correcting heavily damaged images|
|US6711302||Sep 29, 2000||Mar 23, 2004||Eastman Kodak Company||Method and system for altering defects in digital image|
|US6720560||Oct 31, 2000||Apr 13, 2004||Eastman Kodak Company||Method and apparatus for scanning images|
|US6750435||Sep 21, 2001||Jun 15, 2004||Eastman Kodak Company||Lens focusing device, system and method for use with multiple light wavelengths|
|US6862117||Oct 31, 2000||Mar 1, 2005||Eastman Kodak Company||Method and apparatus for reducing the effect of bleed-through on captured images|
|US6924911||Oct 11, 2000||Aug 2, 2005||Eastman Kodak Company||Method and system for multi-sensor signal detection|
|US6987892||Apr 19, 2002||Jan 17, 2006||Eastman Kodak Company||Method, system and software for correcting image defects|
|US7164511||Dec 7, 2000||Jan 16, 2007||Eastman Kodak Company||Distinguishing positive and negative films system and method|
|US8441454 *||Jun 19, 2009||May 14, 2013||Tegic Communications, Inc.||Virtual keyboard system with automatic correction|
|US20020051248 *||Dec 29, 2000||May 2, 2002||Cook Stacy S.||Automatic surface deviation detector and method of use|
|US20020106134 *||Sep 21, 2001||Aug 8, 2002||Dundon Thomas A.||Multiple-orientation image defect detection and correction|
|US20020159165 *||Sep 21, 2001||Oct 31, 2002||Ford Gordon D.||Lens focusing device, system and method for use with multiple light wavelengths|
|US20030011827 *||Dec 7, 2000||Jan 16, 2003||Applied Science Fiction||Distinguishing positive and negative films system and method|
|US20030118249 *||Apr 19, 2002||Jun 26, 2003||Edgar Albert D.||Method, system and software for correcting image defects|
|US20090284471 *||Nov 19, 2009||Tegic Communications, Inc.||Virtual Keyboard System with Automatic Correction|
|DE4143703B4 *||Sep 9, 1991||Jun 19, 2008||Daimler Ag||Visibility improving equipment esp. in motor vehicle - contains optical illumination system operating during successive intervals and receiver system acquiring images of illuminated region during and between illumination intervals for|
|U.S. Classification||348/620, 348/E05.077, 358/447, 382/275|
|International Classification||H04N5/21, H04N7/18, G06T5/00|
|Cooperative Classification||G06T5/50, G06T2207/10016, G06T5/002, H04N5/21|
|European Classification||G06T5/00D, H04N5/21|
|Jan 20, 1987||AS||Assignment|
Owner name: U.S. PHILIPS CORPORATION, 100 EAST 42ND ST., NEW Y
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNORS:HERRMANN, JOCHEM;PELGROM, MARCELLINUS J. M.;REEL/FRAME:004649/0598;SIGNING DATES FROM 19861013 TO 19861114
|Apr 1, 1991||FPAY||Fee payment|
Year of fee payment: 4
|Apr 3, 1995||FPAY||Fee payment|
Year of fee payment: 8
|May 4, 1999||REMI||Maintenance fee reminder mailed|
|Oct 10, 1999||LAPS||Lapse for failure to pay maintenance fees|
|Dec 21, 1999||FP||Expired due to failure to pay maintenance fee|
Effective date: 19991013